Session 5: Projecting the age patterns of mortality, fertility and migration Models and exercises.
Population Aging and Economic Growth in Asia v162.1 The effects of mortality and fertility changes...
Transcript of Population Aging and Economic Growth in Asia v162.1 The effects of mortality and fertility changes...
-
Population Aging and Economic Growth in Asia
David E. Bloom
David Canning
Jocelyn E. Finlay
Harvard School of Public Health
September 2008
DRAFT: Please do not cite without permission from the
authors
Abstract: The decline in the total fertility rate between 1960 and 2005, coupled with an
increase in life expectancy and the dynamic evolution of past variation in birth and death
rates, is producing a significant shift in age structure in Asia. The age distribution has
shifted from one with a high youth-age population share to one with a high old-age
population share. We illustrate the role of these separate forces in shaping the age
distribution. We also argue that the economic consequences of population aging depend
on behavioral responses to the shift in age structure: the female labor force participation
response to the decline in fertility, child quality/quantity trade-off in the face of the
fertility decline, savings adjustments to an increase in life expectancy, and social security
distortions insofar as the pace of life expectancy improvements is faster than the pace of
policy adjustments. We estimate the association between old- and youth-age population
shares and economic growth. The results suggest that population aging may not
significantly impede economic performance in Asia in the long run.
-
2
1. Introduction
The demographic transition and evolution of past birth and death rates in Asia
have brought about dramatic shifts in the age structure between 1960 and 2005 (Bloom,
Canning and Fink 2008). The combined forces of declining fertility, increasing life
expectancy, and the transitional dynamics of varying cohort sizes moving through the age
distribution have led to the rapid aging of societies across much of Asia.1 From 1960 to
2005, China experienced the largest absolute increase in life expectancy in the world.
During the same period, the total fertility rate in the Republic of Korea plummeted from
5.7 to 1.1 – a change that only a handful of countries have experienced. Japan boasts the
highest life expectancy in the world at 85.6 for women and 78.7 for men, and it continues
to rise.
With a decline in fertility, in the short run the youth-age population share declines
and the working-age share increases. Working-age people contribute to the labor force
more than youth-age, and if these individuals are gainfully employed (Bloom, Canning,
Fink and Finlay 2007) then while income per worker can remain the same, income per
capita increases. In Asia, the decline in the total fertility rate from a regional average of
6.05 (see Table 1) in 1960, to a regional average of 2.63 has brought with it an increase in
the working-age share. However, as the total fertility rate falls below the replacement rate
in many Asian countries the working-age share will decrease in the long run (Bloom,
Canning, Fink, Finlay 2008) and old-age shares will increase.
As reviewed in Bloom, Canning and Fink (2008), a popular view of the negative
effects of aging is borne of the growth accounting calculation. If labor supply and savings
behavior remain unchanged, then labor supply, savings (and thus income) per capita
would decline as old-age shares increase and working-age shares decline. However, the
dramatic change in family structure creates avenues for behavioral change. With fewer
children to care for and the support of elderly parents to care for children and contribute
1 Aging in Asia has been particularly dramatic, but it is also taking place in nearly all other regions of the
world.
-
3
to household expenses, individuals who are of working age may be able to work more
than they could previously. Savings patterns may change as life expectancy, income
potential, and expenditure requirements change. Furthermore, incentives to invest in
one’s own and one’s children’s education may change as life expectancy increases and
earning opportunities expand. These behavioral effects may add up to offset the negative
accounting effects of aging.
There are three key drivers of population aging: fertility decline, increase in life
expectancy, and age-structure dynamics. When we factor in the behavioral responses to
the changes in the various demographic variables, we find that these different
demographic forces have different effects on economic growth. The decline in fertility
causes an increase in the female labor supply (Bloom, Canning, Fink and Finlay 2007),
an increase in life expectancy will alter savings incentives (Bloom, Canning and Moore
2007), and a combination of factors leads to increased investment in education per
person. The accounting effects of aging, combined with these behavioral responses, mean
that aging has an ambiguous effect on economic growth.
Changing institutional settings compound the complexity of analyzing the effects
of aging on economic growth. Reforms in social security and diminished adherence to
filial piety make for a transformative situation for identifying the responsible agent for
elder support and care: adult children, the state, own savings or labor supply, or a
company pension.
Will aging have a negative effect on economic growth in Asia? This is the key
question that we explore in this paper. In the next section we break down what we
actually mean by aging and illustrate the role of fertility decline and mortality
improvements on shifts in age structure. In section three we use regression analysis to
identify the statistical relationship between age structure and economic growth in Asia,
and discuss the various behavioral responses to aging in the Asian context. A summary
follows.
-
4
2. Population Aging in Asia
We classify a population with an increasing share of old-age persons as one that is aging.
This shift in the age structure is brought about by a decline in fertility that reduces the
number of youth, and thus with no change in the size of the elderly population, the
elderly share increases. An increase in this share can also come about by a
disproportionate increase in the survival of 65+ individuals relative to the improvement in
survival rates of other age groups. An increasing share of old-age individuals can also
result from past variation in birth and death rates. These three forces all contribute to the
trend in most Asian countries of an increasing proportion of 65+ individuals (Figure 6).
The reason for the fertility decline is a contentious issue debated among
economists and demographers. Bongaarts (1984; 1994; 1999) argues that the
improvements in contraceptive access have aided the decrease in fertility rates. But
Pritchett asserts that access to contraception cannot explain why fertility rates have fallen
by so much, so quickly, and that a shift in preferences explains most of the plummet in
fertility rates (Pritchett 1994). Deciphering why the total fertility rate has fallen in Asia is
not the focus of this paper; we analyze the economic consequences of the observed
decline in the total fertility rate and the associated increase in old-age population shares.
Improvements in life expectancy in many Asian countries between 1960 and 2005
in large part reflect declines in child mortality. Improvements in mortality can be
attributed to public health interventions (for example, improvements in nutrition and the
provision of water and sanitation) and to medical interventions such as vaccine coverage
and the use of antibiotics (Cutler, Deaton and Lleras-Muney 2006). But health disparities
within Asia remain broad: childhood mortality remains high in Laos and Cambodia, for
example; life expectancy is the highest in the world in Japan, and survival to 60 is close
to certain; and adult mortality is high in West Asian countries relative to East Asia.
Despite the disparities, life expectancy has increased in all Asian countries (See Table 1)
between 1960 and 2005.
-
5
Aging is also a consequence of the dynamic evolution of past fertility and
mortality rates. Cohorts move through the age groups of the population age distribution.
The size of the 80-85 age group in 2000 will depend on births in 1915-1920 and the
mortality rates this cohort experienced as it aged. When the total fertility rate falls below
replacement, the birth cohort will be smaller than the parent cohort (excluding migration
effects).
In this section, we examine the effect of fertility and mortality changes between
1960 and 2005 on age structure. This exercise illustrates the dominant role of the fertility
decline in shaping the increase in the proportion of old-age individuals. We illustrate how
age structure will change between 2005 and 2050 as a consequence of fertility and
mortality changes during that period. We also discuss the effects of the dynamic
evolution of 1960-2005 changes in mortality and fertility rates on age structure changes
between 2005 and 2050.
2.1 The effects of mortality and fertility changes on age structure
In Figure 1, we illustrate the effect of fertility and mortality changes on age structure
between 1960 and 2005. As an example of how to interpret these graphs, consider India
and the 0-5 age group. The “fertility effect” of approximately 0.07 indicates that if
fertility rates had remained at 1960 rates then in 2005 the fraction of individuals aged 0-5
years old would be seven percentage points higher. The “mortality effect” of
approximately –0.001 indicates that if age-specific mortality rates had remained at 1960
rates then the fraction of individuals in the 0-5 age group would be 0.1 percentage points
lower than it actually is today. For each of the represented Asian countries (India,
Indonesia, Vietnam, China, Japan, Republic of Korea), we see that the fertility changes
have had a much greater effect on age structure than the mortality changes. In Figure 1,
we see that the fertility decline in the six example countries has had a similar effect on
the respective country’s age structure (though the effect in Japan is minimal). If fertility
rates had remained at 1960 rates, youth population shares would be higher, working-age
-
6
shares would be lower, and old-age shares would be lower2. Figure 1 illustrates the
accounting effect of a decline in fertility leading to an increase in the working-age share
(or as the graph represents: if the fertility rate had not fallen from 1960 rates, the
working-age share would be lower). As to the mortality effect, we see that in four of the
six example countries3 if mortality rates had remained at 1960 rates then the youth
population share would be lower (as more of the children would have died). With the
exception of Vietnam and possibly China, working-age shares would be higher (as
improvements in mortality have been concentrated in the childhood age-groups).
The cohorts born in the 1960-2005 period do not reach the 65+ age group until
2025-2070. Even so, the fertility decline between 1960 and 2005 has an effect on the 65+
age group share. In the case of India in 2005, Figure 1 illustrates that if the total fertility
rate had remained at the 1960 rate then the 65+ age-group population shares would be
lower. Changes in the total fertility rate in the present do not affect old-age population
sizes in the present, but they do affect old-age population shares in the present.
For the age structure of a population to change (with zero migration), either (a)
the fertility rate must change, (b) there must be heterogeneous change in the age-specific
mortality rates, or (c) there must have been past variation in mortality and fertility rates.
As shown in Figure 1, the effect of fertility changes on age structure in the
selected Asian countries has been much more dramatic than mortality changes (even for
China, where female life expectancy climbed from 37.6 to 73.7). As the improvements in
life expectancy are a result of improvements in mortality rates across all of the age
groups, the proportion of individuals in each age group does not change much even in the
face of such steep life expectancy improvements. However, fertility improvements are
concentrated in the 0-1 age group, which leads to an immediate effect on age structure.
2 Old-age shares would be lower in all example countries except for the case of Japan, where the fertility
decline between 1960 and 2005 has been very small relative to the other example countries. 3 Japan is different as in 1960 child mortality rates were already very low compared to the other example
countries.
-
7
A change in fertility or a heterogeneous change in mortality has a cohort effect
that will have an instantaneous effect on age structure. Importantly, this effect will persist
through the age distribution as cohorts move through the different age groups. These
population dynamics imply that changes in the 1960-2005 period will continue to affect
the age structure through the 2005-2050 period (as discussed in Section 2.3).
2.2 The ‘problem’ of aging societies
The ‘problem’ with an aging society is that the number of old-age dependents will
increase relative to the number of working-age individuals. Without deep analysis, it
seems obvious that if people of working age are the workers and people in old age are
retired, and that if changes in the age structure of the population bring about no change in
behavior (an increase in savings, for example), then there will be a rising number of old-
age people dependent on the support of working-age individuals.
We show in Figure 2 that the working-age to dependents ratio will eventually
decline throughout Asia, and comparing Figure 3 and Figure 4 we can see that this will
be due to the declining ratio of working-age people to old-age dependents. In all Asian
regions the ratio of working-age people to old-age dependents will decline (Figure 4);
however, the ratio of working-age to youth-age dependents will increase in all Asian
regions except East Asia (Figure 3). In the past, the rise in the working-age share was
backed by a decrease in youth-age shares. In the future, the decline in the working-age
share will be backed by an increase in the old-age shares.
The major reason population aging matters is that human productivity and human
consumption have different time profiles. Children consume more than they produce.
This phase now lasts into the late 20s in many countries as they continue in advanced
education. Between 25 and roughly 65 are the prime working years, in which production
exceeds consumption. After 65, consumption exceeds labor income. In most cases the
young are supported by intra-family transfers. Support for the elderly, who have normal
consumption and require medical care, is more complex, coming from family support,
-
8
personal savings, pensions, and social security transfers. The mix of these support
systems for the elderly differs greatly across countries. The coping ability of transfer
systems (whether mediated by the family or the state) is limited by the rising proportion
of the elderly and the consequent high burden on the working-age population.
2.3 Aging in the future
In section 2.1 we showed that much of the aging of a society is driven by the decline in
fertility. Sharp declines in the fertility rate decrease the number of youth and increase the
proportion of individuals in old age, even though the population of old people remains
unchanged. Largely as a consequence of the fertility decline, age structure shifted sharply
between 1960 and 2005. Changes in the age structure between 2005 and 2050 also appear
to be steep in Figure 6. However, by comparing Figure 5 with 2005 and 2050 age
structure in Figure 6, we see that changes in age structure between 2005 and 2050 are not
fully consequent on changes over that period – in particular the age structure changes for
the 50+ age groups. This is due to the fact that, in addition to the 2005-2050 fertility and
mortality effects on age structure, shifts in the age structure between 2005 and 2050 will
be a result of changes in cohort sizes stemming from steep changes in fertility and
mortality rates between 1960 and 2005, as these cohorts move through the age
distribution.
As the fertility rate is already below replacement in many Asian countries, and the
mortality improvements are diminishing, changes in age structure as a result of changes
in fertility and mortality changes over the next 45 years will not be as dramatic as they
were between 1960 and 20054. With fertility and mortality rates stabilizing at low levels,
the changes in the age structure over the next 45 years will largely reflect dynamic
evolution of past birth and death rates, and the age structure will move toward stability.
To illustrate the dynamic evolution effects of past birth and death rates on the age
structure, consider the case of Indonesia in Figure 5. The graph indicates that if fertility
4 India is an exception to this.
-
9
and mortality rates remain at 2005 rates, then by 2050 the share of 65-70 year olds would
be 0.5 percentage points less than what the predicted 2050 65-70 year old age share is, as
represented in Figure 6. From Figure 6, we see that the share of 65-70 year olds increased
by 0.035 (3.5 percent); 0.005 of that is explained by fertility and mortality changes
between 2005 and 2050, but the remainder (0.03) is explained by past variation in birth
and death rates. Thus aging, the increase in the proportion of 65+ individuals, between
2005 and 2050 is largely consequent on fertility and mortality rate changes prior to 2005.
3 Population Aging and Economic Growth in Asia
Asia’s, and in particular East Asia’s, macroeconomic performance is tracked very closely
by its demographic transition and resulting changes in age structure. Estimates indicate
that as much as one-third of its “economic miracle” can be attributed to a demographic
dividend (Bloom and Williamson 1998; Bloom, Canning and Malaney 2000; Bloom,
Canning and Sevilla 2001; Bloom, Craig and Malaney 2001; Bloom, Canning and Sevilla
2003). The first demographic dividend comes about as an accounting effect of a decline
in fertility and the resultant rise in the working-age share. When the fertility rate declines,
the working-age share increases as the number of individuals of working age increases
relative to the number in the youth age groups. With an increase in the working-age
share, countries stand to benefit from the proportional increase in the pool of potential
workers in the economy, and income per capita can increase. By contrast, the absence of
demographic change also accounts for a large portion of Africa’s economic debacle
(Bloom and Sachs 1998; Bloom, Canning et al. 2003). In addition, the introduction of
demographic factors has reduced the need for the argument that there was something
exceptional about East Asia or idiosyncratic to Africa. Most models of economic growth
have significant region dummies, usually negative for Sub-Saharan Africa and positive
for East Asia, indicating that the poor performance of Africa and the exceptionally good
growth performance of East Asia cannot be explained within the models. Once age
structure dynamics are introduced into an economic growth model, these regions are
much closer to reflecting widely understood drivers of economic growth (Bloom,
-
10
Canning et al. 2000) and the statistical significance of the region dummy variables
disappears.
The effect of changes in the working-age share on economic growth is well
documented. We now turn our focus to the effect of old-age share on economic growth.
This analysis is not independent of the relationship between working-age share and
economic growth. However, when considering the working-age share affecting economic
growth, the downward forces of youth- and old-age shares is considered symmetric. In
this paper, we treat youth- and old-age shares separately so as not to impose this
symmetry assumption.
3.1 Channels by which aging affects economic growth
In Part 2 of this paper we illustrated in detail the process of aging in Asia. Sharp declines
in fertility between 1960 and 2005 had dramatic effects on the age structure of the
population. Moreover, we showed how the age structure will continue to evolve as
population cohorts age between now and 2050.
In this section, we analyze the effect of this shift in age structure on economic
growth, with a particular focus on Asia. There are many channels by which the shift in
age structure affects economic growth. In the first instance we illustrate the empirical
relationship between age-structure shifts and economic growth. Secondly, we discuss the
behavioral response to a rapid shift in the age distribution: household-level life-cycle
decisions may be influenced by society-level age structure composition. The third
channel we examine is the role of institutional settings in the face of rapid changes in age
structure. In particular, we look at the role of old-age social security and the incentives
created by slow-changing laws.
-
11
3.2 Economic Growth and Age Structure
When analyzing the effects of age structure on economic growth, typically the working-
age share is isolated as the age group of interest. In the coming years, however, concern
over the potentially depressing effects of old-age dependency on economic growth and
the differing effects of youth- and old-age dependency has increased interest in directly
observing the partial effect of these latter variables on economic growth.
We take the convergence model framework outlined in Bloom and Canning
(2008), and dissect population into youth, C , working age, WA , and old, O .
Growth of GDP per worker is characterized by the distance from steady state,
( )* 0zg z zλ= − (1)
and a vector of variables, X , can affect the steady state level of labor productivity.
Following the discrete time models in Barro and Sala-i-Martin (2004), there is a log-
linearization around the steady state. Thus logY
zL
=
, and the steady state level of
income per worker is summarized as,
( )0zg X zλ β= − (2)
To then utilize this theory of convergence in an income per capita model, consider the
relationship used in Bloom and Canning (2008) that highlights the inclusion of age
structure in the form of working-age share,
Y Y L WA
N L WA N= (3)
If age structure is represented by working-age share, and the participation rate is constant,
then,
/ / /Y N Y L WA Ng g g= + (4)
Growth of income per worker is explained by the convergence term in (2). As for the
growth of the working-age share, we wish to observe the separate effects of youth- and
-
12
old-age dependency on the growth of income per capita and not just the growth of the
working-age share. Thus we want to identify,
( )/ / /Y N Y L N C O Ng g g − −= + (5)
To isolate the growth of the youth- and old-age we can draw on rules of approximation.
Firstly, the difference in logs is an approximation for the growth of the working-age
share,
/0
ln lnWA N
t
WA WAg
N N
≈ −
(6)
Then, the working-age is the population less the youth- and old-age populations, so
growth of the working age can be redefined as,
( ) /0
ln lnN C O N
t
N C O N C Og
N N− −
− − − − = −
(7)
The approximation accuracy of ( )ln 1 x x− ≈ is increasing in the working-age share, and
we use this approximation to assert that,
( )
( )
/
0
/0 0
N C O N
t
N C O Nt t
C O C Og
N N
C C O Og
N N N N
− −
− −
+ + = − +
= − + −
(8)
Using the following substitutions,
log , log , logY Y WA
y z wN L N
= = =
(9)
And the fact that taking the logs of both sides of (3) gives,
y z w= + (10)
Then,
-
13
( )
( )
/ 00 0
/ 0 00 0
/ 0 00 0 0 0
/ 0 0 1 2 3 0 4 50 0
Y Nt t
Y Nt t
Y Nt t
Y Nt t
C C O Og X z
N N N N
C C O Og X w y
N N N N
C O C C O Og X y
N N N N N N
C O C Og X y
N N N N
λ β
λ β
λ β
β β β β β β
= − + − + −
= + − + − + −
= − − − − − − −
= + + + + ∆ + ∆
(11)
Thus the effects of youth- and old-age dependency on growth of income per capita can be
estimated using the specification outlined in (11).
An implication of equation (5) is that the coefficients on the change in youth- and
old-age shares are minus one. Any deviation from this coefficient is brought about by
misspecification. That is, the youth-age and old-age will have heterogeneous effects on
income per capita as changes in the youth- and old-age shares incite different behavioral
responses. High youth-age dependency will cause women to exit the workforce to care
for children. Higher old-age shares, in part brought about by higher life expectancies,
may reflect lower morbidity thus extending the time which people stay in the workforce.
Moreover, the older individuals may accumulate capital (we control for this in our
regression analysis) for a longer period before they draw down on savings.
Included in the vector of explanatory variables describing the steady state level of
GDP per worker, X, are capital stock, education, and institutional quality, and other
factors that may affect labor productivity. We also include a global time trend and a
random error term in this vector. To measure the growth of income per capita we take the
difference in logs. Thus, the estimated equation is,
( ) ( )1 1 1 2 1 3 1 4 1 5 1 6 1
6 6
ln lnt t t t t t t t
t t i t it
y y y edu cap inst c o
c o
β β β β β β
β β δ δ ε
− − − − − − −− = + + + + +
∆ + ∆ + + + (12)
Fixed effects with a lagged dependent variable introduce a Nickell (1981) bias.
Thus, we do not estimate a fixed effects model and instead we control for some potential
-
14
country fixed effects with time invariant country specific geographic variables such as
fraction of land in the tropics, and whether the country is landlocked or not. We also
control for regional dummies.
It is reasonable to be concerned about possible endogeneity of the age structure
variables in the economic growth equation. Periods of high growth may increase the
working-age share through migration effects, or have an effect on fertility and thus the
youth-age population shares (Bloom and Canning 2008). To control for this issue of
reverse causality, we use five-year lags of the age structure variables (and an alternative
instrument set using five-year-lag fertility and life expectancy).
Using data from the Penn World Tables 6.2 (Heston, Summers and Aten 2006)
for GDP per capita and Penn World Tables 5.6 for capital stock, and demographic data
from the World Population Prospects (United Nations 2007), we estimate the effect of
changes in age structure on economic growth. In Table 5 we detail the data sources with
citations.
In Table 2 we present the descriptive statistics of all of the variables, including the
instrumental variables, in the regression analysis. We see from these statistics that a range
of countries are represented in the sample: high to low income, low to high fertility rates,
low to high infant mortality rates. In Table 3 we detail the country lists of the country
dummies used. The sample of countries is larger than this, and the country lists only
detail those of the continent dummies included.
In Table 4, we present the regression results. In column one, we present the
ordinary least squares (OLS) results for the long run demographic model. In column two
we introduce the short run effects of youth- and old-age share changes. For these latter
variables, a change in the youth- or old-age share will have an impact in the short run on
the annual average five year economic growth rate. The initial level of the youth- and
old-age shares will affect the long run growth rate. In column three we present the two-
stage-least-squares results for the demographic model. We use the five year lag of the
-
15
youth- and old-age change in population share, and the initial year fertility rate and initial
year infant mortality rate. In columns four and five we introduce the Asian and African
continent dummies.
We see from the regression results that changes in both youth- and old-age shares
over a 5 year period have a negative effect on the short-run (5 year) growth rate. This
effect is negative and significant even in the two-stage-least-squares specifications in
columns three and five. Once we control for the continent dummies, however, the results
indicate that the long run effect of the level of old-age population share is negative but
not significant. The long run effect of the level of the youth-age population share is,
however, negative and significant.
Capital stock, trade openness, and institutional quality each have a positive and
significant effect on the five year growth rate. The significance of these control variables
is consistent with other studies: even when controlling for these core variables that
explain cross country differences in economic growth, the level and change in the age
structure of the population have an effect on economic growth. Highlighting the
importance of demographic change in explaining economic growth is a feature of Bloom,
Canning and Sevilla (2003) but in this paper, we treat youth-age and old-age population
shares as heterogeneous. The positive and significant East Asian dummy indicates that on
average over 1960-2005 period East Asia’s economic growth rate was higher than the
global average.
Taking the results in column 5 of Table 4, we can interpret the magnitude of the
coefficients on the demographic variables. A 10 percentage point decrease in the youth-
age share, will increase the economic growth rate by 2.2 percentage points leading to a
higher steady state income per capita in the long run. From the results, we see that a
change in the old-age share does not have a significant effect on economic performance
in the long run. In the short run, if the change in the youth-age share decreases by one
percent point over a 5 year period, then the average annual economic growth rate will
increase by 0.7 percentage points. If the change in the old-age share increase by one
-
16
percentage point then the average annual economic growth rate in that five year period
will decline by 1.5 percentage points. Thus increasing old-age shares we observe
throughout Asia will have a negative effect on economic growth, but in the long run the
rising share of elderly may not impede economic performance.
In past studies, the positive influence of an increase in working-age share may, in
part, have been a proxy for the decline in youth-age dependents. The decline in the
working-age share that is to come in East Asia over the next 45 years is coupled with an
increase in old-age, and not youth-age, dependency. The observed partial effect of
changes in the growth of the working-age share on economic growth may differ in the
future, as declines in the working-age share will be coupled with increases in the old-age
share and not increases in the youth-age share. From the regression results in Table 4,
however, we see that for East Asia, the heterogeneity of the effect of youth- and old-age
shares on economic growth is evident in the long run, but not in the short run. In the short
run, changes in either youth- and old-age shares have a similar magnitude of effect on
economic growth. However, the long run effect differs: in the long run low youth-age
shares will have a positive effect on economic growth, but high old-age shares may have
an insignificant effect on economic growth.
Old-age dependency is not a given. The compression of morbidity means that a
significant portion of this population can continue to work, if they so desire. By working
longer they can save more than in the past for their retirement. But rising income means
greater demand for leisure and retirement, and this effect may dominate, leading to early
retirement (Costa 1995). However, if retirement is voluntary and older people have saved
enough for their old age, they are not dependents. Dependency means financing old-age
consumption through transfers. If retirement is funded by productive savings or own
labor supply, then the elderly should not be considered 'dependent'. The case of youth
dependency is clearer. Born into the world with no financial assets, the young do not
have the savings or work-effort potential that the elderly have, and they are clearly
dependent.
-
17
Our results suggest that high youth-age shares are negatively associated with
economic performance in the long run, and high old-age shares may not have a
significant effect the long run steady state level of income per capita. This stands to be
potentially good news for those Asian countries that are about to experience a decline in
working-age share coupled with an increase in the old-age shares. The increase in income
in the past, in part boosted by the increase in the working-age share, was backed by the
decline in the youth-age shares. In the future, the decline in working-age share will not
necessarily bring with it a decline in income, as this demographic trend will be backed by
an increase in old-age population shares and not an increase in youth-age shares.
3.3 Behavioral change: labor supply, savings, and education
Although the model we proposed in the above section implies age structure variables
have coefficients of –1 in the growth equation, the estimation revealed otherwise. This
indicates that there is some kind of measurement error in age structure effects. This
measurement error can be explained by accounting and behavioral responses to age
structure shifts.
Given well-established life-cycle variations in behavior, it is reasonable to
suppose that changes in age structure will have effects on aggregate outcomes. Changes
in age structure bring with them changes in labor supply, savings, and education as the
number of people engaging in the various life-cycle decisions changes. For example,
since labor supply tends to follow an inverted U-shaped pattern with respect to age,
changes in the age composition of the population are likely to have effects on aggregate
labor supply. Savings rates also vary with age, with the highest rates occurring for 40- to
70-year olds, implying that changes in the age structure will affect aggregate savings
rates. Furthermore, increases in life expectancy mean that more people survive through
the school ages, and the average number of years of education increases.
However, in addition to these accounting effects there are also behavioral effects
of aging. Generational crowding (i.e., being born into a large cohort) may have effects on
-
18
relative wages and individual labor supply (Easterlin 1980; Bloom, Freeman and
Korenman 1988; Korenman and Neumark 2000). In addition, falling fertility and youth
dependency rates may be linked to increased labor market participation, particularly
among women, as found in Bloom, Canning, Fink and Finlay (2007).
Improvements in life expectancy (which is a proxy for better health) that are
inherent to an aging society can invoke behavioral responses that have a positive effect
on economic growth. Better health may improve worker productivity (Bloom, Canning
and Sevilla 2004). However, there may also be a demographic effect as a longer
prospective lifespan can change life-cycle behavior, leading to a longer working life and
higher saving for retirement (Bloom, Canning and Moore 2005; Bloom, Canning,
Mansfield and Moore 2007; Bloom, Canning et al. 2007). Moreover, a higher life
expectancy may increase the incentive to invest in education, as the years over which
returns can be amortized are extended (Finlay 2006).
3.3.1 Labor supply
In Bloom, Canning, Fink and Finlay (2007), the authors show that the decline in total
fertility rate has had a significant effect on the increase in female labor force
participation. They show that a reduction in the fertility rate of one child is associated
with an increase in labor force participation of four years. With fewer children, women
have more opportunities to stay in, or re-enter, the workforce as the time required by
child care declines. Increased child care services, and the decline in the stigma of a
working mother, have helped to make the option of women staying in (or re-entering) the
workforce more attractive.
In Bloom, Canning, Fink and Finlay (2007), the Republic of Korea is used as an
example to illustrate the effect of fertility decline on economic growth. The authors show
that in Korea demographic effects explain about 14 percent of the increase in income per
capita. The decline in population growth, the increase in the working-age share, and then
-
19
the positive female labor force participation response to the decline in the fertility rate all
contribute to the rapid rise in income per capita in the Republic of Korea.
The behavioral female labor supply response contributes about 25 percent of the
14 percent increase in income per capita. This is only one of the behavioral responses that
can occur during the demographic transition. Further analysis of the savings and
education responses may find compounded effects compared with those found in Bloom,
Canning, Fink and Finlay (2007) and may thus yield even higher estimates of the income
effects of demographic change.
3.3.2 Savings
Central to our understanding of the East Asian "miracle" has been Alwyn Young's
work (1994; 1995) showing that rapid economic growth in the region was mainly due to
increases in factor inputs––notably labor, capital, and education––and not to
improvements in total factor productivity5. In order to understand the rise in income
levels in East Asia we must therefore understand the driving forces behind the growth in
these inputs. All of the Asian "Tiger" economies enjoyed a surge in savings and
investment during their period of rapid economic growth. We focus here on Taiwan, for
which there are fairly good data on household savings. The private savings rate in Taiwan
rose from around 5% in the 1950s to well over 20% in the 1980s and 1990s. Savings
rates vary by age, being highest in Taiwan for households with heads in the 50-60 year
old range. We would therefore expect changing age structure to be a possible explanation
for this increase in aggregate saving. Studies that examine the link between demographic
structure and national savings rates do find a strong connection (Leff 1969; Fry and
Mason 1982; Mason 1987; Mason 1988; Kelley and Schmidt 1995; Kelley and Schmidt
1996; Higgins and Williamson 1997; Higgins 1998) and suggest that a large part of the
savings boom in East Asia can be explained by the changing age structure of the
population.
5 This argument is controversial, as discussed in Krugman, P., (1994), The myth of Asia’s economic
miracle, Foreign Affairs 73, 62-78.
-
20
However, Deaton and Paxson (2000) show that, based on household saving data
for Taiwan, changes in age structure account for only a modest increase in the overall
savings rate, perhaps 4 percentage points. They show that the rise in the aggregate
savings rate has not been mainly due to changes in the age composition of the population
but, rather, to a secular rise in the savings rates of all age groups.
The question then arises as to why savings rates rose at each age. One possible
explanation, proposed by Lee, Mason, and Miller (2000) is that increased savings rates
are due to rising life expectancy and a subsequent need to fund retirement income. Tsai,
Chu, and Chung (2000) show that the timing of the rise in household savings rates
matches the increases in life expectancy of the population.
With a fixed retirement age we would expect such a savings effect. However,
Deaton and Paxson (2000) argue that in a flexible economy, without mandatory
retirement, the main effect of a rise in longevity will be on the span of the working life,
with no obvious prediction for the rate of saving. Bloom, Canning, and Moore (2005)
formalize this argument to show that under reasonable assumptions the optimal response
to an improvement in health and a rise in life expectancy is to increase the length of
working life, though less than proportionately, with no need to raise saving rates at all
(due to the gains from enjoying compound interest over a longer life span).
The effect of savings on investment and domestic production depends on the
nature of the capital market. With perfect capital mobility, demographic change may have
an impact on international capital flows (Higgins 1998). In this case, effects on domestic
interest rates and investment may be minimal (Poterba 2004). However, if capital markets
are imperfect the demographic transition can lead to a mismatch between the investment
needs of a large, young, working-age population and the retirement savings of older
workers (Higgins and Williamson 1997).
-
21
3.3.3 Education
Demography can affect educational investments through several mechanisms.
Perhaps the most important is the quantity-quality tradeoff whereby fertility choices and
human capital investment decisions are jointly made. This framework points to lower
fertility rates being both a cause and a consequence of increased educational investments,
with both fertility and schooling determined by a common set of factors that affect
families’ incentives.
Notwithstanding families’ desired fertility, actual fertility in the absence of
contraception may be much higher. The provision of family planning services to
populations in which desired fertility is low can both lower fertility outcomes and
increase schooling levels. This effect may be particularly pronounced for girls’ schooling
because with high fertility, girls are frequently kept out of school to help care for their
younger siblings. Foster and Roy (1997) show how a randomized trial providing family
planning services in Bangladesh affected both fertility outcomes and children's schooling
levels.
The quantity-quality tradeoff can also appear to some extent at the national level
if schooling is publicly funded. Smaller youth cohorts can increase the availability of
educational funding per child and can lead to an expansion of public education (Kelley
1996; Lee and Mason 2008).
One reason for an increased incentive to invest in education may be the rise in life
expectancy. A longer life increases the time over which education investments can be
recouped. Kalemli-Ozcan, Ryder and Weil (2000) argue that the effect of improved
health and longevity on educational investments has played a large role in economic
growth over the last 150 years. This incentive effect, however, is clearly linked to the
prospective working life rather than total lifespan, suggesting that education levels may
be linked to planned retirement ages and social security incentives.
-
22
3.4 Institutional settings: social security
With health improvements and longer life expectancies, the optimal response for
workers with perfect markets may be to have a longer working life. However, mandatory
or conventional retirement ages, coupled with the strong financial incentives to retire that
are inherent in many social security systems, seem to result in early retirement and
increased needs for saving for old age (Bloom, Canning et al. 2007).
Generous state transfer systems not only have financing problems, they
undermine and reduce labor supply of the elderly, increasing effective dependency rates.
Many social security systems impose a very high effective tax rate on older workers by
withholding or reducing benefits if they continue to work.
Singapore, Malaysia, and Hong Kong (China) have fully funded universal
systems. These systems consist of personal accounts so an older worker who continues
working benefits from a larger sum to retire on. These systems should not discourage
work at older ages and should be associated with high savings rates. Taiwan, China,
India, Vietnam, and Indonesia do not have universal systems. In these countries planning
for retirement has historically been rare. However, they do have systems for the formal
sector and public sector that can generate large future liabilities.
Specific social security systems were designed for existing demographic
situations and may not be appropriate as the proportion of elderly continues to rise.
However, transforming these systems once established is very difficult politically, as
entitlements under the systems are difficult to reduce. In countries without universal
systems, population aging will put pressure on governments to provide more coverage,
given the difficulties experienced by families trying to cope with the issue. The systems
put in place will have a large impact on how aging affects those economies.
-
23
4 Summary
In this paper we have illustrated the effects of aging on economic growth in Asia. Aging
is driven by a decline in fertility, an increase in life expectancy, or the dynamic evolution
of past variation in birth and death rates. In the first part of the paper we illustrate these
driving forces of aging. We show that between 1960 and 2005 the shift in age structure of
the population has been predominantly driven by the rapid decline in fertility, and
between 2005 and 2050 the dynamic evolution of fertility and mortality changes in the
1960-2005 period will continue to shape the age distribution. In the second part of the
paper, we estimated the statistical relationship between the youth-age, and old-age
population shares and economic growth, both in the long run and in the short run.
Regression analysis indicates that in the long run old-age shares may not have a
significant impact on economic growth. A change in the old-age shares has a negative
effect on economic growth in the short run, but not in the long run. The level of youth-
age shares (and the change) has a negative effect on long run economic performance.
This result is good news for Asian countries in the long run as the old-age population
share increases.
We discuss that population aging has more than a simple accounting effect on
economic growth and we discuss the various behavioral responses that come with the
shift in age structure: an increase in female labor force participation as fertility declines,
an increase in savings, and an increase in education. These factors act together and may
offset any negative accounting effects of a shift in the age structure, thus in the regression
analysis we observe an insignificant effect of rising old-age shares on economic
performance in the long run. Overall, the effect of aging on economic growth will be
ambiguous, as the various behavioral responses may impose economic growth effects of
differing magnitudes across different countries.
-
24
5 References
Barro, R. J. and J.-W. Lee (2000). "International Data on Educational Attainment:
Updates and Implications." CID Working Paper 42.
Barro, R. J. and X. Sala-i-Martin (2004). Economic Growth, MIT Press.
Bloom, D. E. and D. Canning (2008). "Global Demographic Change: Dimensions and
Economic Significance." Population and Development Review 34(Supplement):
17-51.
Bloom, D. E., D. Canning and G. Fink (2008). "Population Aging and Economic
Growth." PGDA working paper 31.
Bloom, D. E., D. Canning, G. Fink and J. E. Finlay (2007). " Fertility, Female Labor
Force Participation, and the Demographic Dividend." NBER Working Paper
13583.
Bloom, D. E., D. Canning, G. Fink and J. E. Finlay (2007). "Realizing the Demographic
Dividend: Is Africa any Different?" PGDA working paper 23.
Bloom, D. E., D. Canning and P. Malaney (2000). "Population Dynamics and Economic
Growth in Asia." Population and Development Review 26(supp.): 257-90.
Bloom, D. E., D. Canning, R. K. Mansfield and M. Moore (2007). "Demographic
Change, Social Security Systems and Savings." Journal of Monetary Economics
54: 92-114.
Bloom, D. E., D. Canning and M. Moore (2005). "The Effect of Improvements in Health
and Longevity on Optimal Retirement and Savings." PGDA working paper 2.
Bloom, D. E., D. Canning and M. Moore (2007). "A Theory of Retirement." PGDA
working paper 26.
Bloom, D. E., D. Canning and J. Sevilla (2001). "Economic Growth and the
Demographic Transition." NBER Working Paper(8685).
Bloom, D. E., D. Canning and J. Sevilla (2003). "The Demographic Dividend: A New
Perspective on the Economic Consequences of Population Change." Population
Matters Monograph MR-1274, RAND, Santa Monica.
Bloom, D. E., D. Canning and J. Sevilla (2004). "The Effect of Health on Economic
Growth: A Production Function Approach." World Development 32(1): 1-13.
Bloom, D. E., P. H. Craig and P. N. Malaney (2001). The Quality of Life in Rural Asia.
Hong Kong, Oxford University Press.
Bloom, D. E., R. Freeman and S. Korenman (1988). "The labor market consequences of
generational crowding." European Journal of Population 3(2): 131-176.
Bloom, D. E. and J. Sachs (1998). "Geography, demography, and economic growth in
Africa." Brookings papers on Economic Activity 2: 207-273.
Bloom, D. E. and J. G. Williamson (1998). "Demographic transitions and economic
miracles in emerging Asia." World Bank Economic Review 12(3): 419-455.
Bongaarts, J. (1984). "Implications for future fertility trends for contraceptive practice."
Population and Development Review 10(2): 341-352.
Bongaarts, J. (1994). "The Impact of Population Policies: Comment." Population and
Development Review 20(3): 616-620.
-
25
Bongaarts, J. (1999). "Fertility Decline in the Developed World: Where Will It End?"
The American Economic Review. Papers and Proceedings of the One Hundred
Eleventh Annual Meeting of the American Economic Association 89(2): 256-260.
Costa, D. (1995). "Pensions and Retirement: Evidence from Union Army Veterans."
Quarterly Journal of Economics 110(2): 297-320.
Cutler, D., A. Deaton and A. Lleras-Muney (2006). "The Determinants of Mortality."
Journal of Economic Perspectives 20(3): 97-120.
Deaton, A. and C. Paxson (2000). "Growth, Demographic Structure, and National
Savings in Taiwan." Population and Development Review 26(supp.): 141-173.
Easterlin, R. (1980). Birth and fortune: the impact of numbers on personal welfare. New
York, Basic Books.
Foster, A. and N. Roy (1997). The dynamics of education and fertility: evidence from a
family planning experiment.
Fry, M. and A. Mason (1982). "The Variable Rate-of-Growth Effect in the Life-cycle
Saving Model." Economic Inquiry XX: 426-43.
Heston, A., R. Summers and B. Aten (2006). "Penn World Table Version 6.2." Center for
International Comparisons of Production, Income and Prices at the University of
Pennsylvania.
Higgins, M. (1998). "Demography, National Savings and International Capital Flows."
International Economic Review 39(2): 343 - 369.
Higgins, M. and J. G. Williamson (1997). "Age Structure Dynamics in Asia and
Dependence on Foreign Capital." Population and Development Review 23(2):
261-93.
Kalemli-Ozcan, S., H. E. Ryder and D. N. Weil (2000). "Mortality decline, human capital
investment, and economic growth." Journal of Development Economics 62(1): 1-
23.
Kelley, A. C. (1996). The consequences of population growth for human resource
development: the case of education. The impact of population growth on well-
being in developing countries. D. A. Ahlburg, A. C. Kelley and K. Oppenheim-
Mason. Berlin, Springer-Verlag: 67-137.
Kelley, A. C. and R. M. Schmidt (1995). "Aggregate Population and Economic Growth
Correlations: The Role of the Components of Demographic Change."
Demography 32(4): 543-555.
Kelley, A. C. and R. M. Schmidt (1996). "Saving, dependency, and development."
Journal of Population Economics 9: 365-386.
Korenman, S. and D. Neumark (2000). Cohort crowding and youth labor markets: a
cross-national analysis. Youth employment and joblessness in advanced countries.
D. G. Blanchflower and R. B. Freeman, University of Chicago/NBER: 57-105.
Lee, R. and A. Mason (2008). "Fertility, Human Capital, and Economic Growth over the
Demographic Transition." Mimeo.
Lee, R., A. Mason and T. Miller (2000). "Life Cycle Saving and Demographic
Transition: The Case of Taiwan." Population and Development Review 26(supp.):
194-219.
Leff, N. H. (1969). "Dependency Rates and Savings Rates." American Economic Review
59(5): 886-96.
-
26
Mason, A. (1987). National Saving Rates and Population Growth: A New Model and
New Evidence. Consequences of Population Growth in Developing Countries. G.
Johnson and R. D. Lee. Madison, Wisconsin, University of Wisconsin Press: 523-
560.
Mason, A. (1988). Population Growth, Aggregate Saving and Economic Development.
World Population Trends and their Impact on Economic Development. D.
Salvatore. London, Greenwood Press: 45-58.
Nickell, S. (1981). "Biases in dynamic models with fixed effects." Econometrica 49(6):
1417-1426.
Poterba, J. (2004). "The Impact of Population Aging on Financial Markets." NBER
Working Paper 10851.
Pritchett, L. (1994). "Desired Fertility and the Impact of Population Policies." Population
and Development Review 20(1): 1-55.
Sala-i-Martin, X., G. Doppelhofer and R. I. Miller (2004). "Determinants of Long-Term
Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach."
American Economic Review 94(4): 813-35.
Tsai, I.-J., C. Y. C. Chu and C.-F. Chung (2000). "Demographic Transition and
Household Savings in Taiwan." Population and Development Review 26(supp.):
174-193.
United Nations (2007). World Population Prospects: The 2006 Revision. CD-ROM
Edition - Extended Dataset United Nations.
World Bank (2007). World Bank Development Indicators CD-ROM.
Young, A. (1994). "Lessons from the East Asian NICs: A Contrarian View." European
Economic Review 38: 964-973.
Young, A. (1995). "The Tyranny of Numbers: Confronting the Statistical Realities of the
East Asian Growth Experience." Quarterly Journal of Economics 110: 641-680.
-
27
6 Appendices
Table 1: Summary Statistics
1960 2005 1960 2005 1960 2005 1960 2005
World (average) 3,821 9,887 5.51 3.03 55.22 68.60 1.32 1.68
World (standard deviation) 3,567 10,221 1.78 1.65 12.71 13.37 0.30 0.43
Asia (average) 2,254 10,529 6.05 2.63 52.89 72.67 1.24 1.80
Asia (standard deviation) 1,720 11,118 1.34 1.23 10.99 7.04 0.17 0.45
Afghanistan . 581 7.70 . 34.4 . 1.20 1.03
Armenia . 4,765 4.47 1.37 69.2 76.4 1.30 2.14
Azerbaijan . 3,671 5.55 2.33 68.2 75.1 1.33 2.09
Bahrain . 19,561 7.09 2.34 57.5 76.3 1.03 1.91
Bangladesh . 2,155 6.81 2.98 39.9 64.8 1.16 1.55
Bhutan . 933 5.90 2.50 38.5 65.2 1.32 1.32
Brunei Darussalam . 26,239 6.83 2.38 62.9 79.4 1.11 2.07
Cambodia . 580 6.29 3.89 44.1 60.6 1.21 1.52
China 445 5,333 3.39 1.81 37.6 73.7 1.26 2.42
Cyprus . 22,383 3.44 1.42 70.6 81.7 1.41 2.13
Georgia . 5,788 2.95 1.39 67.5 74.9 1.75 1.97
China, Hong Kong Special Administrative Region 3,264 29,644 5.06 0.97 69.8 84.5 1.25 2.91
India 870 2,990 6.57 2.84 43.5 64.3 1.31 1.65
Indonesia 1,099 4,064 5.52 2.27 42.3 69.7 1.32 1.95
Iran (Islamic Republic of) 3,269 6,398 7.00 2.07 48.5 72.8 1.09 2.04
Iraq . 1,230 7.27 . 51.2 . 1.05 1.28
Israel 6,526 21,242 3.87 2.82 73.2 81.8 1.46 1.61
Japan 4,632 24,660 2.00 1.26 70.1 85.6 1.82 1.81
Jordan 4,187 3,742 7.75 3.29 48.5 73.6 1.09 1.43
Kazakhstan . 10,169 . 1.75 . 71.9 1.48 2.10
Democratic People's Republic of Korea . 1,429 4.37 1.96 56.8 66.9 1.52 2.09
Republic of Korea 1,544 18,421 5.67 1.08 55.8 81.1 1.24 2.46
Kuwait . 26,098 7.27 2.39 61.5 79.7 1.06 2.16
Kyrgyzstan . . . 2.41 . 72.4 1.37 1.65
Lao People's Democratic Republic . 1,412 6.15 4.50 41.8 57 1.23 1.27
Lebanon . 6,085 5.70 2.25 63 74.8 1.16 1.83
China, Macao Special Administrative Region . 37,956 5.02 0.88 61.9 82.4 1.20 3.14
Malaysia 1,829 12,131 6.81 2.74 55.9 76.1 1.03 1.70
Maldives . 5,086 7.00 4.00 42.6 67.3 1.19 1.24
Mongolia . 1,597 6.00 2.33 48.3 68.5 1.18 1.93
Myanmar . . 6.00 2.23 45.4 64.1 1.25 1.92
Nepal 818 1,441 6.06 3.46 38.4 63.1 1.29 1.40
Oman . 16,273 7.20 3.44 42.9 76.3 1.08 1.43
Pakistan 803 2,685 6.92 4.12 43.1 65.5 1.20 1.36
Philippines 2,037 3,938 6.96 3.20 55.3 73.2 1.05 1.57
Qatar . 36,183 6.97 2.89 55 76.6 1.22 1.93
Saudi Arabia . 16,010 7.22 3.83 45.8 74.6 1.14 1.34
Singapore 4,211 29,419 5.45 1.24 65.7 81.6 1.14 2.54
Sri Lanka 865 4,274 5.35 1.91 56.8 77.4 1.15 2.11
Syrian Arab Republic 829 2,016 7.48 3.24 51 75.7 1.03 1.51
Tajikistan . 1,942 6.26 3.53 58.9 66.7 1.29 1.36
Thailand 1,086 7,275 6.40 1.89 58.3 74.5 1.11 2.24
Democratic Republic of Timor-Leste . . 6.36 7.47 34.6 57.8 1.23 1.23
Turkey 2,264 5,982 6.28 2.19 52.1 73.8 1.18 1.86
Turkmenistan . 7,342 6.46 2.60 58.3 67.2 1.31 1.74
United Arab Emirates . 35,676 6.91 2.43 55 81.6 1.15 1.84
Uzbekistan . 3,915 . 2.22 62.7 70.7 1.32 1.64
Viet Nam . 2,561 6.05 1.78 46.3 73.2 1.28 1.87
Yemen . 1,076 8.36 5.87 34.9 63.2 1.05 1.06
GDP per capita Total Fertility Rate Life Expectancy (female) WA/dependents
Source: Penn World Tables 6.2 (Heston, Summers et al. 2006); World Development
Indicators 2007 (World Bank 2007)
-
28
Figure 1: The influence of fertility and mortality change on age structure shifts 1960-
2005
-.0
50
.05
.1.1
5D
iffe
ren
ce
0 20 40 60 80age_group
Fertility Effect Mortality Effect
India 1960-2005
-.0
50
.05
.1.1
5D
iffe
ren
ce
0 20 40 60 80age_group
Fertility Effect Mortality Effect
Indonesia 1960-2005
-.0
50
.05
.1.1
5D
iffe
ren
ce
0 20 40 60 80age_group
Fertility Effect Mortality Effect
Vietnam 1960-2005
-.0
50
.05
.1.1
5D
iffe
ren
ce
0 20 40 60 80age_group
Fertility Effect Mortality Effect
China 1960-2005
-.0
50
.05
.1.1
5D
iffe
ren
ce
0 20 40 60 80age_group
Fertility Effect Mortality Effect
Japan 1960-2005
-.0
50
.05
.1.1
5D
iffe
ren
ce
0 20 40 60 80age_group
Fertility Effect Mortality Effect
Korea 1960-2005
Source: Authors' own calculations using World Population Prospects (United Nations
2007), World Development Indicators (World Bank 2007), and ModMatch in Stata 9.
Figure 2: Ratio of Working-age Population to Dependent Population, by Asian Region
-
29
11.5
22.5
1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050year
Eastern South Central
South East Western
Working-age to Dependent Population
Source: World Population Prospects (United Nations 2007)
-
30
Figure 3: Ratio of Working-age Population to Youth Population, by Asian Region
12
34
5
1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050year
Eastern South Central
South East Western
Working-age to Youth Population
Source: World Population Prospects (United Nations 2007)
-
31
Figure 4: Ratio of Working-age Population to Old-age Population, by Asian Region
05
10
15
20
1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050year
Eastern South Central
South East Western
Working-age to Old Population
Source: World Population Prospects (United Nations 2007)
-
32
Figure 5: The Influence of Fertility and Mortality Change on Age Structure Shifts 2005-
2050
-.0
2-.
01
0.0
1.0
2.0
3D
iffe
ren
ce
0 20 40 60 80age group
Fertility Effect Mortality Effect
India 2005-2050
-.0
2-.
01
0.0
1.0
2.0
3D
iffe
ren
ce
0 20 40 60 80age group
Fertility Effect Mortality Effect
Indonesia 2005-2050
-.0
2-.
01
0.0
1.0
2.0
3D
iffe
ren
ce
0 20 40 60 80age group
Fertility Effect Mortality Effect
Vietnam 2005-2050
-.0
2-.0
10
.01
.02
.03
Diffe
ren
ce
0 20 40 60 80age group
Fertility Effect Mortality Effect
China 2005-2050
-.0
2-.0
10
.01
.02
.03
Diffe
ren
ce
0 20 40 60 80age group
Fertility Effect Mortality Effect
Japan 2005-2050
-.0
2-.0
10
.01
.02
.03
Diffe
ren
ce
0 20 40 60 80age group
Fertility Effect Mortality Effect
Korea 2005-2050
-
33
Figure 6: Age structure 1960, 2005, 2050. The proportion of individuals in the 5-year age
group (0 on the x-axis is the 0-4 age group, 5 is the 5-9 age group). 0
.05
.1.1
5.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100
India
1960 2005 2050
0.0
5.1
.15
.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100
Indonesia
1960 2005 2050
0.0
5.1
.15
.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100
Vietnam
1960 2005 2050
0.0
5.1
.15
.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100
China
1960 2005 2050
0.0
5.1
.15
.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100
Japan
1960 2005 2050
0.0
5.1
.15
.2
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100
Korea
1960 2005 2050
Source: World Population Prospects (United Nations 2007)
-
34
Table 2: Descriptive Statistics
Variable Mean Std.Dev. Min Max
Five year growth rate of income per capita 1.35 3.36 -23.48 20.38
Log real GDP per capita 8.44 1.15 5.14 10.49
Capital stock 0.63 1.96 0.00 22.88
Average years of secondary schooling 5.45 2.83 0.35 12.05
Trade openness 65.23 46.14 3.78 462.93Freedom House Polity Index 6.14 3.30 0.19 10.00
Life Expectancy 64.06 11.84 31.20 82.10
Tropical Location 0.53 0.48 0.00 1.00
Land Locked 0.18 0.38 0.00 1.00
Youth-age share 0.35 0.10 0.14 0.51
Old-age share 0.06 0.05 0.01 0.20
Youth-age share change -0.01 0.01 -0.07 0.02
Old-age share change 0.003 0.005 -0.012 0.026Total Fertility Rate 3.88 1.96 1.08 8.50
Infant Mortality Rate 53 46 3 199
N=616
-
35
Table 3: Country Dummy Lists
East Asia South Central Asia South East Asia Sub-saharan Africa Latin America
China Bangladesh Indonesia Benin Argentina
Japan India Malaysia Botswana Bolivia
Republic of Korea Iran (Islamic Republic of) Philippines Cameroon Brazil
Nepal Singapore Central African Republic Chile
Pakistan Thailand Democratic Republic of the Congo Colombia
Sri Lanka Congo Costa Rica
Gambia Dominican Republic
Ghana Ecuador
Guinea-Bissau El Salvador
Kenya Guatemala
Lesotho Haiti
Liberia Honduras
Malawi Jamaica
Mali Mexico
Mauritius Nicaragua
Mozambique Panama
Niger Paraguay
Rwanda Peru
Senegal Trinidad and Tobago
Sierra Leone Uruguay
South Africa Venezuela
Sudan
United Republic of Tanzania
Togo
Uganda
Zambia
Zimbabwe
-
36
Table 4: Regression Results
(1) (2) (3) (4) (5)
ols ols iv ols iv
Log real GDP per capita in the base year -2.184*** -2.077*** -2.153*** -2.037*** -2.111***
(0.35) (0.34) (0.36) (0.39) (0.41)
Capital stock in the base year 0.140*** 0.168*** 0.171*** 0.112** 0.120***
(0.041) (0.040) (0.044) (0.045) (0.041)
Average years of secondary schooling in the base year -0.0114 -0.0577 -0.0246 -0.0909 -0.0641
(0.080) (0.079) (0.081) (0.078) (0.081)
Trade openness in the base year 0.0137*** 0.0143*** 0.0110*** 0.0149*** 0.0105***
(0.0032) (0.0030) (0.0025) (0.0033) (0.0027)
Freedom House Polity Index in the base year 0.152*** 0.140*** 0.155*** 0.167*** 0.187***
(0.050) (0.048) (0.051) (0.051) (0.054)
Life Expectancy in the base year 0.146*** 0.0943*** 0.0748** 0.0606** 0.0497
(0.027) (0.028) (0.032) (0.030) (0.034)
Youth-age share in the base year -19.53*** -21.44*** -24.68*** -18.37*** -21.95***
(3.30) (3.57) (4.41) (3.95) (4.62)
Old-age share in the base year -29.46*** -18.38*** -18.76** -8.902 -10.87
(6.93) (6.77) (7.52) (7.36) (8.19)
Youth-age share change -57.73*** -75.70*** -53.57*** -70.38***
(10.5) (14.8) (10.8) (14.3)
Old-age share change -57.18** -125.2*** -64.44** -149.5***
(23.0) (46.6) (25.0) (48.9)
Tropical location -1.103*** -0.928*** -0.879** -1.104** -1.135**
(0.35) (0.33) (0.35) (0.43) (0.44)
Landlocked 0.389 0.506 0.692* 0.606* 0.754*
(0.37) (0.37) (0.39) (0.37) (0.39)
Latin America Dummy 0.300 0.259
(0.44) (0.46)
Sub-saharan Africa -0.773 -0.568
(0.56) (0.59)
East Asia Dummy 2.038*** 1.895***
(0.63) (0.63)
South Central Asia Dummy -0.307 -0.389
(0.59) (0.59)
South East Asia Dummy 1.305** 1.554***
(0.58) (0.58)
Constant 20.40*** 22.78*** 21.58*** 19.14*** 21.80***
(3.15) (3.24) (3.80) (4.14) (4.58)
Year Dummies Yes Yes Yes Yes Yes
Observations 722 720 649 714 643
R-squared 0.27 0.31 0.26 0.32 0.28
Cragg-Donald F-stat 45.70 40.51
Hansen J p-value 0.411 0.500
Robust standard errors in parentheses; *** p
-
37
Table 5: Data Sources
Variable Source
Five year growth rate of income per capita Penn World Tables 6.2, (log(Real GDP per capita)(t) - log(Real GDP per capita)(t-5))/5*100
Log real GDP per capita Penn World Tables 6.2, log(Real GDP per capita) PPP constant 2000 prices
Capital stock Penn World Tables 5.6, interpolated
Average years of secondary schooling Barro and Lee (2000), Average years of schooling for people ages 15 years of older
Trade openness Penn World Tables 6.2, Exports plus Imports divided by GDP per capita, constant pricesFreedom House Polity Index http://www.freedomhouse.org/template.cfm?page=1, degree of democracy less degree of autocracy
Life Expectancy World Development Indicators 2007, Life Expectancy at Birth
Tropical Location Sala-i-Martin, Doppelhofer, Miller (2004), Proportion of land in tropical area
Land Locked Sala-i-Martin, Doppelhofer, Miller (2004), Landlocked country dummy
Youth-age share World Development Indicators 2007, Fraction of Population 0-14 years old
Old-age share World Development Indicators 2007, Fraction of Population 65+ years old
Youth-age share change World Development Indicators 2007, Youth age share (t) - Youth age share (t-1)
Old-age share change World Development Indicators 2007, Old age share (t) - Old age share (t-1)Total Fertility Rate World Development Indicators 2007, Total Fertility Rate
Infant Mortality Rate World Development Indicators 2007, Infant (